This invention relates generally to processing within a computing environment, and more particularly to error detection in a computing environment.
A commonly used base ten numeric representation is a packed decimal format referred to as binary coded decimal (BCD). In BCD format, a number is represented in a series of eight-bit bytes, where four bits of every byte are allocated to a single decimal digit, allowing each byte to represent two decimal digits.
Another base ten numeric representation is a densely packed decimal (DPD) format. In DPD format, three decimal digits are represented in ten bits.
Some contemporary processors execute decimal floating point operations on decimal floating point numbers that have a sign bit, a combination field, an exponent continuation field, and a coefficient continuation field. The coefficient continuation field may be compressed using a DPD encoding format where three decimal digits that normally take up twelve bits in a BCD encoding format are compacted into ten bits. However, the DPD encoded data must be converted into BCD encoded data before carrying out computations on the processor.
Thus, there is often a requirement to convert from BCD to DPD (may be referred to as BCD to DPD compression) and to convert from DPD to BCD (may be referred to as DPD to BCD expansion). At least two types of error checking may be part of the conversion process. Error checking may be performed on the input data (e.g., via a parity bit send with the input data) to verify that the expected data was received at the converter. In addition, error checking may be performed on the results of the conversion to verify the results of the conversion process.